* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Analysis of process-induced residual stresses in tape placement
Carbon nanotubes in interconnects wikipedia , lookup
Fracture mechanics wikipedia , lookup
Creep (deformation) wikipedia , lookup
Glass transition wikipedia , lookup
Shape-memory alloy wikipedia , lookup
Thermal copper pillar bump wikipedia , lookup
Viscoplasticity wikipedia , lookup
Thermal expansion wikipedia , lookup
Strengthening mechanisms of materials wikipedia , lookup
Stress (mechanics) wikipedia , lookup
Cauchy stress tensor wikipedia , lookup
Work hardening wikipedia , lookup
Lumped element model wikipedia , lookup
Paleostress inversion wikipedia , lookup
Journal of Thermoplastic Composite Materials http://jtc.sagepub.com Analysis of Process-Induced Residual Stresses in Tape Placement Fazil O. Sonmez, H. Thomas Hahn and Mustafa Akbulut Journal of Thermoplastic Composite Materials 2002; 15; 525 DOI: 10.1177/0892705702015006207 The online version of this article can be found at: http://jtc.sagepub.com/cgi/content/abstract/15/6/525 Published by: http://www.sagepublications.com Additional services and information for Journal of Thermoplastic Composite Materials can be found at: Email Alerts: http://jtc.sagepub.com/cgi/alerts Subscriptions: http://jtc.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations (this article cites 18 articles hosted on the SAGE Journals Online and HighWire Press platforms): http://jtc.sagepub.com/cgi/content/refs/15/6/525 Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. Analysis of Process-Induced Residual Stresses in Tape Placement FAZIL O. SONMEZ,1,* H. THOMAS HAHN2 AND MUSTAFA AKBULUT1 1 Department of Mechanical Engineering, Bogazici University Istanbul, Bebek 80815 Turkiye 2 Mechanical and Aerospace Engineering Department UCLA, Los Angeles CA 90024, USA ABSTRACT: The tape placement for thermoplastic composites involves heating, melting, and cooling steps just as do the other manufacturing processes. Consequently, development of residual stresses is unavoidable due to disparate thermal characteristics of matrix and fiber materials and also due to nonuniform cooling rates. From the product quality standpoint, such as interlaminar strength, dimensional accuracy etc., these stresses should be kept within allowable limits. In this study, a thermoviscoelastic finite element model was developed to predict residual stresses induced during the placement of thermoplastic composite tapes. The process, being continuous, was considered to be under a quasi steady state where process conditions do not change with respect to the moving roller. Relaxation of the residual stresses in previously laid layers was also allowed for. Results were obtained for both unidirectional and cross-ply laminates. They show the residual stress distributions through the thickness for a number of chosen sets of process parameters (e.g., roller velocity and heat input). Therefore, residual stresses in a laminate can be controlled by modifying these process parameters. KEY WORDS: thermoplastic composite laminates, APC-2, residual stresses, finite element analysis (FEA). INTRODUCTION inevitably involves residual stress development. Residual stresses can lead to distortion of finished components, matrix cracking, and inter-ply delaminations [1,2]. Tensile P ROCESSING OF THERMOPLASTIC composites *Author to whom correspondence should be addressed. E-mail: [email protected] Journal of THERMOPLASTIC COMPOSITE MATERIALS, Vol. 15—November 2002 525 0892-7057/02/06 0525–20 $10.00/0 DOI: 10.1106/089270502023207 ß 2002 Sage Publications Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. 526 FAZIL O. SONMEZ, H. THOMAS HAHN AND MUSTAFA AKBULUT stresses may even reach to a significant fraction of the tensile strength. Therefore, to ensure reliable and satisfactory performance, residual stresses within a composite part should be known before the part is put into use. Unacceptably high levels of stress may arise because of two factors. First of all, thermal expansion behavior of fiber-reinforced thermoplastic composites is highly anisotropic. This is due to the large discrepancy in the thermal expansion coefficients of the matrix and fiber materials. In a typical carbon-reinforced unidirectional thermoplastic composite, a temperature increase induces considerable expansion of the laminate in the transverse direction but almost none along the fiber direction. Besides, fibers, being themselves anisotropic, contribute to this effect. The other factor is the temperature gradients induced during the cooling process. This is because of the fact that, when different regions of the composite experience different temperature histories and stiffness properties of the material vary with the temperature, deformation behavior will be incompatible. Although elastic properties of fibers are almost unaffected by temperature, influence of temperature on the properties of the polymer matrix is drastic. Also high processing temperatures typical for high performance thermoplastic composites exacerbate both the effect of temperature gradients and the effect of anisotropic thermal expansion behavior. The tape placement process is one of the few techniques that have the potential to continuously process thermoplastic composites in large scale industrial production. In the process, an incoming composite tape is bonded to the previously laid and consolidated laminate under heat and pressure locally applied to the interface (Figure 1). By laying additional layers in different directions, a part with desired thickness and properties can be fabricated. Since heat is mainly supplied to the region where the tape and the laminate meet, the process is highly nonisothermal and, residual stresses may become critical. The process poses a very challenging problem of determining residual stresses for the following reason: The tape placement is a continuous process. As such, when a new layer is being placed, the previously laid and consolidated layers are again subjected to heating. If the temperature of these layers exceeds the glass transition temperature, annealing will ensue and the residual stresses developed during previous tape placement will relax. Most of the previous studies on the prediction of residual stresses developed during thermoplastic composite processing concentrated on press molding [2–12]. On the other hand, Nejhad et al. [13] proposed a model to predict process induced stresses in filament winding. They modeled filament winding as a transient process. Since this led to excessive computational time, residual stresses could not be obtained within a reasonable time. As for Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. Analysis of Process-Induced Residual Stresses in Tape Placement 527 Figure 1. The thermoplastic composite tape placement process. Figure 2. The solution procedure. the modeling of residual stresses in tape placement, the present study is, as far as the authors know, the first of its kind. Figure 2 shows the solution procedure used in the analysis. The thermal and stress analyses are coupled. The temperature field generated by the thermal analysis is used in the stress analysis to calculate thermal strains and Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. 528 FAZIL O. SONMEZ, H. THOMAS HAHN AND MUSTAFA AKBULUT shift factors. Since a different kind of mesh structure is utilized in the stress analysis, the temperature field determined by the thermal analysis is mapped onto the grid used in the stress analysis by a linear interpolation. In order to account for the relaxation of residual stresses in previously laid layers, residual stresses calculated during the placement of the last ply are used as initial stresses for the current configuration. HEAT TRANSFER ANALYSIS The previously developed model for heat transfer in the tape placement process [14] was used, but generalized to account for cross-ply lay-ups. STRESS ANALYSIS The residual stress model used in the present study is also an extension of the model that was developed to predict instantaneous stress distribution under the roller [15]. The model was generalized to account for cross-ply lay-ups. Because the objective in this study was to determine residual stresses after the completion of tape placement process rather than the instantaneous stress field under the roller, the mesh structure was changed. Instead of a mesh whose elements are concentrated under the roller, a mesh having lengthwise uniform elements is used. A vertical shift factor was introduced for the creep compliance. Temperature increase not only accelerates the relaxation process but also increases long term compliance (S1 ), which has considerable effect on the residual stress level. Besides, the original finite element formulation was modified to account for the residual stresses developed during the placement of the previous layers. Formulation of the Problem ASSUMPTIONS Fabrication of 3D structures having smooth surfaces is possible in the tape placement process. Tapes can be placed even on nongeodesic paths. But, in this study a basic geometry was chosen to simplify the analysis and interpretation of its results. Specifically, the laminate was assumed to be flat with a unidirectional or a cross-ply lay-up. Although, in practice, the roller should be narrow to accommodate curved contours, it was assumed to be as wide as the laminate itself in the present study, so that a state of plane strain could be used (Figure 3). The effect of crystallization on residual stress development was assumed to be negligible as some previous studies [6,12] had found it to be. The roller was assumed to move with a constant Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. Analysis of Process-Induced Residual Stresses in Tape Placement 529 Figure 3. Analyzed configuration of the process. velocity vx. A steady state was thus assumed to exist within the Eulerian control volume moving with the roller away from an edge. CONSTITUTIVE RELATIONS In the analysis, linear constitutive relations were used. Displacement gradients ui, j are much smaller than unity. In several studies [3,6], curvature of unbalanced APC-2 laminates was used as a measure of residual stresses during thermal processing. From these studies, it may be concluded that a significant portion of the residual stresses builds up between just above the glass transition temperature (Tg) and the room temperature. Time dependent response of the material was reported to be dominated by the viscoelastic process close to Tg [16]. Nonlinear and viscoplastic effects can therefore be neglected without compromising the accuracy of the results. For a continuous unidirectional fiber-reinforced composite having a viscoelastic matrix, the creep compliances S11 and S12 can be taken to be independent of time and temperature [16,17], where ‘‘1’’ denotes the fiber direction and ‘‘2’’ denotes the transverse direction. The plane of the transverse isotropy is 2–3 plane, and the roller moves in the x direction. The constitutive relations appropriate for this case are expressed as [18–20] "x ðtÞ "*x ðtÞ ¼ S11 x ðtÞ þ S12 y ðtÞ þ S12 z ðtÞ Zt @y ðÞ d V22 ðÞS22 ð 0 Þ "y ðtÞ "*y ðtÞ ¼ S12 x ðtÞ þ @ 0 Zt @z ðÞ þ d V23 ðÞS23 ð 0 Þ @ 0 Zt @xy ðÞ d 2"xy ðtÞ ¼ V66 ðÞS66 ð 0 Þ @ 0 Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. ð1Þ 530 FAZIL O. SONMEZ, H. THOMAS HAHN AND MUSTAFA AKBULUT where "*x ðtÞ and "*y ðtÞ are the thermal strains. Note that a state of plane strain with "z ¼ 0 is assumed. Since the thermal expansion coefficients, i, are temperature dependent, they are given by "*i ðtÞ ¼ Z Tf i ðTÞ dT ð2Þ To The symbols and 0 are the pseudo times, which represent the dependence of the creep compliance on temperature. They are given by Z t ¼ a½TðÞ d and 0 0 ¼ Z a½TðÞ d ð3Þ 0 a(T ) is the horizontal shift factor required to shift the modulus–time curve at temperature T to the one at the base temperature Tbase. When T is less than Tbase, a(T ) is smaller than 1, and vice versa. This means that accelerated viscoelastic processes within the material at a higher temperature are reflected in a faster elapse of time. V(T ) is the vertical shift factor introduced to account for the increase in long term creep compliance S(1) with the increasing in temperature. When the roller moves normal to the fibers, the resulting constitutive relations are Z t @x ðÞ d V22 ðTÞS22 ð 0 Þ @ 0 Zt @y ðÞ þ d þ S12 z ðtÞ V23 ðTÞS23 ð 0 Þ @ 0 Zt @x ðÞ V23 ðTÞS23 ð 0 Þ d "y ðtÞ "*y ðtÞ ¼ @ 0 Zt @y ðÞ d þ S12 z ðtÞ V22 ðTÞS22 ð 0 Þ þ @ 0 Zt dxy ðÞ d ½S22 ð 0 Þ S23 ð 0 Þ 2"xy ðtÞ ¼ d 0 "x ðtÞ "*x ðtÞ ¼ ð4Þ BOUNDARY CONDITIONS The laminate rests on a rigid mold. Therefore, at the bottom surface, the following boundary condition is met: uy ¼ 0 Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. ð5Þ Analysis of Process-Induced Residual Stresses in Tape Placement 531 At the left side of the control volume, the initial stress ( x) is equal to the residual stress developed during the placement of the previous ply: x ðx, yÞx¼0 ¼ i1 ð yÞ ð6Þ The control volume is sufficiently large so that the stress gradients are zero at the right and left sides: @ ¼0 @x ð7Þ This condition can be achieved, only if the temperature gradients are also equal to zero. This means the control volume should be taken so large that the temperature at the right edge is the same as the ambient temperature. With the condition of constant temperature, stresses also remain constant. Therefore, the stress state at the right edge of the control volume represents residual stresses. Because we assume linear constitutive relations, namely a relation which is independent of the level of stress, the force applied by the roller on the laminate is not included in the analysis. The roller induces instantaneous stresses which quickly decay to zero as soon as the roller passes. Therefore, we assume that the roller force has no effect on the development of residual stresses. Finite Element Solution of the Problem Since the roller moves with a constant velocity, we use the assumption of a quasi-steady state neglecting the inertia effect. Of course, such assumption does not hold true near the edges of the part. Thus, the results of the present analysis are valid away from the edges. The quasi-steady state assumption permits the use of a control volume which moves with the roller. The control volume is taken as stationary and the material body moves through the control volume. The control volume to be analyzed is divided into eight-degrees-of-freedom rectangular elements as shown in Figure 4. The material body thus moves through the control volume to the right. The strain in a finite element is a linear function of the eight nodal displacements [21]: f"gr ¼ ½Ar fgr ð8Þ where [A]r is a 3 8 matrix whose elements depend only on the nodal displacements of the element r, {}r. Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. 532 FAZIL O. SONMEZ, H. THOMAS HAHN AND MUSTAFA AKBULUT Figure 4. Mesh structure used in the stress analysis. Horizontal and vertical dimensions are not to scale. Consider the material element which occupies the control volume element n at time tn. We note that this material element was at the control volume element 1 at time t1, at the control volume element 2 at time t2, etc., Figure 4. The viscoelastic memory dependence of this material element implies that the stress state in the control volume element n at tn be determined by the strains in all elements 1, 2, . . . , ( n 1) which the material element passed through previously [22,23]. Thus, the stress components in the finite element n can be found by converting the convolution integrals in Equation (1) into a summation of series and reducing the relation to the following form (Appendix A): 9 8 n9 2 nr 38 r r 0 > c11 cnr 0 > = X = < x > < "x "*x þ "x > n 12 r r 0 nr 4 cnr 5 * yn ¼ " " þ " c 0 ð9Þ y y y 12 22 > > > : n > ; r¼1 ; r nr : 2" 0 0 c xy xy 66 or fgn ¼ n X r¼1 ½Cnr f"gr n X ½Cnr f"*gr þ r¼1 n X ½Cnr f"0 g ð10Þ r¼1 where f"*gr is the thermal strain vector [15] whose elements depend only on the nodal temperatures of the element r, and f"o g is the initial strain vector representing residual stresses developed during the placement of previous layers. Combining Equations (8) and (10), we obtain fgn ¼ n X r¼1 ½Cnr ½Ar fgr n X ½Cnr f"*gr þ r¼1 n X ½Cnr f"0 g ð11Þ r¼1 The virtual work principle requires that Z f"gTn fgn dVn fgTn fFgn ¼ Vn Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. ð12Þ 533 Analysis of Process-Induced Residual Stresses in Tape Placement where {F}n is the vector of forces acting at the nodal points of the element n, and Vn is its volume. Substituting Equations (8) and (11) into Equation (12) yields " # Z n n n X X X T T T fgn ½An ½Cnr ½Ar fgr ½Cnr f"*gr þ ½Cnr f"0 g dVn fgn fFgn ¼ Vn r¼1 r¼1 r¼1 ð13Þ Therefore, Z n X fFgn ¼ Vn r¼1 Z þ ½ATn ½Cnr ½Ar fgr Z dVn n X Vn r¼1 ½ATn ½Cnr f"*gr dVn n X ½ATn ½Cnr f"0 g dVn ð14Þ Vn r¼1 or, fFgn ¼ n n n X X X ½Knr fgr þ fHgnr þ fH0 g r¼1 r¼1 ð15Þ r¼1 Here [K ]nr is the effective element stiffness matrix, {H }nr is the effective thermal force vector, and {H0} is the effective initial force vector. Components of the element stiffness matrix and the thermal force vector are given in [24]. By following the procedure in [15], the global stiffness equation for the entire control volume can be obtained from the element stiffness equations (Equation (15)), and solved to determine the stress field within the control volume. The stress state of the material exiting the control volume can then be taken to be the same as the residual stress developed after the placement of the layer. RESULTS AND DISCUSSIONS A computer code was developed following the aforementioned solution procedure and the effects of the processing parameters such as roller speed on the residual stress distribution within the laminate were investigated. The processed material was chosen to be APC-2. Inputs for the FEM Code Most of the inputs for the heat transfer and stress analysis were given in [14,15]. Xiao [16] provided the data regarding the creep compliances and the Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. 534 FAZIL O. SONMEZ, H. THOMAS HAHN AND MUSTAFA AKBULUT Table 1. Vertical shift factors for APC-2. Temperature ( C) 40 129.4 200 Vertical Shift Factor (log V ) 0.025 þ0.0031 þ0.21 shift factors. Vertical shift factors reported by Xiao [16] were extrapolated, as was done in [7,25] (Table 1). Barnes et al. [26,27] characterized the anisotropic thermal expansion behavior of APC-2. Jeronimidis and Parkyn [3] determined the stress free temperature by heating up unbalanced laminates and observing the temperature at which they became flat. They found it to be 310 C. Another study [6] found the stress free temperature to be 280 C. However, the residual stress model was observed to be not very sensitive to the value of stress free temperature. The results did not show appreciable differences between the two values of stress free temperature. In the present study, therefore, the base temperature, Tbase, was taken to be 280 C and the effect of thermal expansion on residual stress development was neglected above this temperature by taking thermal expansion coefficients as zero. In order to ensure convergence, the length of the control volume was chosen to be 4 m, and the finite element mesh to be 120 24. Verification Due to the lack of experimental data on the tape placement process, our model was verified using the results of the press molding process. The numerical results were compared with experimental data for two cases. One is a unidirectional laminate, [040], processed at 35 C/s surface cooling rate, and the other a cross-ply laminate, [010/9010]s processes at 0.25 C/s surface cooling rate. As seen in Figures 5 and 6, the residual stress model agrees quite well with the experimental data. Because the modified version of the residual stress model for press molding was verified, we assumed that predictions of the residual stress model developed for tape placement were reliable. The Effect of Process Variables The thermomechanical history determines the resulting residual stress in a viscoelastic material. Therefore, in the present study we chose the process variables that influence temperature distribution and examined their effects Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. Analysis of Process-Induced Residual Stresses in Tape Placement 535 Figure 5. Transverse normal residual stress distribution within a 40 ply APC-2 unidirectional laminate, [040]T, processed by press molding at a 35 C/s surface cooling rate. Experimental data were obtained by a layer removal technique [28]. Figure 6. Normal stress distribution within a cross-ply laminate, [010/9010]s, processed by press molding at a 0.25 C/s surface cooling rate. Experimental data were obtained by a layer removal technique [11]. on residual stress. Figure 7 shows the process variables under consideration. The default values are 16 mm/s for roller velocity (vx), 1.0 for the ratio of heated substrate length to heated tape length ( ¼ hls/hl ). The hot gas temperature is taken as 600 C. The results show the maximum tensile residual stress at the end of 16-ply lay-up, [04/904]s. Preheating the laminate to a certain temperature before placing the tape is a usual practice to help bonding. If the preheating temperature is increased, the temperature difference between the heated zone and the remaining region Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. 536 FAZIL O. SONMEZ, H. THOMAS HAHN AND MUSTAFA AKBULUT Figure 7. Process variables of tape placement. Figure 8. The effect of preheat temperature on the maximum tensile stress, x. becomes smaller, and the material cools more slowly [14]. Consequently residual stresses decrease as Figure 8 indicates. Figure 9 shows increasing roller velocity leading to a higher level of residual stress. This is because, at high speeds, only regions close to the heated surface are raised to high temperatures [14]. Localized heating causes higher temperature gradients, and thus higher residual stresses. As our previous study [14] showed, the lengths of the heated surfaces on the tape and substrate significantly affect the temperature distribution. The effect of larger heated lengths is similar to the effect of preheating the substrate in slowing down the cooling rate. In contrast, a small heated length results in a highly localized heated zone and thus higher residual stresses as suggested by Figure 10. In the above results, both the tape and the substrate were heated equally. Supplying more heat to the substrate by increasing the surface exposed to hot gas was found to result in a more uniform temperature distribution [14]. Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. Analysis of Process-Induced Residual Stresses in Tape Placement 537 Figure 9. The effect of roller velocity on the maximum tensile stress, x. Figure 10. The effect of heated length on the maximum tensile stresses. Figure 11 indeed shows that increasing the ratio of heated substrate length to heated tape length reduces the maximum tensile residual stress. In these runs, the heated length on tape was kept constant at 1.5 cm while the heated length on substrate was increased from 1.5 to 4.5 cm, then to 7.5 cm, and finally to 10.5 cm. Figure 12 shows residual stress ( xx) distribution through the thickness of a cross-ply laminate, [04/904]s. Ply numbers 1 and 16 indicate the top and bottom layers, respectively. The distribution is uneven through the thickness of the laminate. Compressive stresses develop in the top and bottom 4 plies Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. 538 FAZIL O. SONMEZ, H. THOMAS HAHN AND MUSTAFA AKBULUT Figure 11. The effect of heated length ratio on the maximum tensile stress. Figure 12. Residual stress ( xx ) distribution through the thickness of a 16 ply cross-ply laminate. while tensile stresses develop in the middle 8 plies. These stresses are high enough to cause undesirable distortions in the laminate. The tensile stresses occurring in the middle plies are high enough to cause matrix cracking. CONCLUSIONS A residual stress analysis was carried out for the tape placement process, and the effects of process parameters on residual stress distributions were Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. Analysis of Process-Induced Residual Stresses in Tape Placement 539 investigated through a parametric study. The process parameters that affect the temperature distribution were found to affect the residual stresses. Any changes in process parameters leading to a more localized temperature distribution lead to higher residual stresses. The chosen sets of process parameters resulted in quite high residual stresses, much higher than the levels encountered in press molding. Residual stresses accumulate gradually during successive lay-down of layers, they may reach excessively high levels especially in thick laminates. Furthermore, unsymmetrical distribution of residual stresses can lead to the distortion of finished products. Thus, process parameters should be optimized to minimize residual stresses and, at the same time, reduce the uneven distribution. This can be achieved by combining the present process model with an optimization algorithm. Otherwise, post-processing may be needed. In the present investigation we only examined the effects of process parameters on residual stresses. We have not considered other criteria for product quality such as consolidation, crystallinity or degradation. A full optimization of process parameters should include these other quality criteria as well. ACKNOWLEDGMENT This paper is based on the work supported by the Research Fund of Bogazici University with the code number 97HA602. APPENDIX A Numerical Integration of the Constitutive Equations Rewriting constitutive equation for shear strain given in Equation (1), we have Z t V66 ðTðÞÞS66 ððtÞ 0 ðÞÞ 2"xy ðtÞ ¼ 0 @xy ðÞ d @ ðA:1Þ where pseudo-time, , is related to the real time by Z t a½Tðxi , Þ d ¼ ðA:2Þ 0 where a(T ) is the horizontal shift factor. The above integral can easily be calculated numerically. Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. 540 FAZIL O. SONMEZ, H. THOMAS HAHN AND MUSTAFA AKBULUT Applying integration by parts to Equation (A.1), one obtains t 2"xy ðtÞ ¼ xy ðÞ V66 ðTðÞÞS66 ððtÞ 0 ðÞÞ0 Zt @½V66 ðTðÞÞS66 ððtÞ 0 ðÞÞ d xy ðÞ @ 0 ðA:3Þ 2"xy ðtÞ ¼ xy ðtÞV66 ðVðtÞÞS66 ð0Þ xy ð0ÞVðTð0ÞÞS66 ððtÞÞ Zt @½V66 ðTðÞÞS66 ððtÞ 0 ðÞÞ d xy ðÞ @ 0 ðA:4Þ then, where the second term represents the initial stress. Equation (A.4) can be integrated using finite differences in time. Consider incremental times ti or i(ti) (i ¼ 1, 2, . . . , r, . . . , n), where tr is the time at which the material is in the control volume element r (Figure 4), (t1 ¼ 0 and tn ¼ t). Utilizing the second mean value theorem, we can express the strain in the control volume element n as 2"xy ðtn Þ ¼ xy ðtn ÞV66 ðTn ÞS66 ð0Þ xy ðt1 ÞV66 ðT1 ÞS66 ððtn ÞÞ n1 1X xy ðtrþ1 Þ þ xy ðtr Þ 2 r¼1 ½V66 ðTrþ1 ÞS66 ðn rþ1 Þ V66 ðTr ÞS66 ðn r Þ ðA:5Þ Thus Equation (1) can be reduced to the following form: 9 8 0 9 9 9 8 8 2 nr nr 38 " > *x ðtn Þ > > ðt Þ ðt Þ " d d 0 " x n x r > > > > > 11 12 n = > = X = < < x > = < < 6 7 nr nr "y ðtn Þ d22 0 5 y ðtr Þ þ "*y ðtn Þ "0y ¼ ðA:6Þ 4 d12 > > > > > > ; > ; r¼1 ; > : > : > nr : ; : 0 2"xy ðtn Þ xy ðtr Þ 0 0 d66 0 2"xy or f"gn ¼ n X ½Dnr fgr þ f"*gn f"0 g r¼1 where nr ¼ d11 0 S11 r 6¼ n r¼n Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. ðA:7Þ 541 Analysis of Process-Induced Residual Stresses in Tape Placement nr d12 nr ¼ d22 ¼ 0 S12 r 6¼ n r¼n ðA:8Þ 8 1 2fV22 ðT1 ÞS22 ðn Þ þ V22 ðT2 ÞS22 ðn 2 Þg > > > > > < 1fV22 ðTr1 ÞS22 ðn r1 Þ V22 ðTrþ1 ÞS22 ðn rþ1 Þg 2 > V22 ðT1 ÞS22 ð1 Þ > > > > :1 2fV22 ðTn1 ÞS22 ð1 Þ þ V22 ðTn ÞS22 ðn n1 Þg r ¼ 1, n 6¼ 1 1<r<n r¼n¼1 r ¼ n 6¼ 1 ðA:9Þ nr d66 ¼ 8 1 fV66 ðT1 ÞS66 ðn Þ þ V66 ðT2 ÞS66 ðn 2 Þg > > > 2 > > < 1fV66 ðTr1 ÞS66 ðn r1 Þ V66 ðTrþ1 ÞS66 ðn rþ1 Þg 2 > V66 ðT1 ÞS66 ð1 Þ > > > > :1 2fV66 ðTn1 ÞS66 ð1 Þ þ V66 ðTn ÞS66 ðn n1 Þg r ¼ 1, n 6¼ 1 1<r<n r¼n¼1 r ¼ n 6¼ 1 ðA:10Þ Here, 1 is the fiber direction, and 2 and 3 are the transverse directions. Initial strains are related to initial stresses by 8 0 9 2 38 0 9 S11 ð0Þ S12 ð0Þ 0 > > = = < "x > < x > 6 7 0 "0y ¼ 4 S12 ð0Þ V22 ðT1 ÞS22 ððtn ÞÞ ðA:11Þ 0 5 y > > : 0 > ; : 0 > ; 0 0 V ðT ÞS ððt ÞÞ 2"xy 2xy 66 1 66 n Strain–stress relations given by Equations (A.6) and (A.11) in numerical form are valid for plane stress. But in our model, we assumed plane-strain state. For plane strain, when the directions of roller movement, x, and the fiber direction, 1, coincide, (A.6) becomes 9 8 2 nr nr 2 nr nr nr nr nr "x ðtn Þ > d11 ðd12 Þ =d12 d12 d12 d23 =d22 > > > n 6 = X < 6 d nr d nr d nr =d nr d nr ðd nr Þ2 =d nr "y ðtn Þ ¼ 4 12 12 23 22 22 23 22 > > > > ; r1 : 2"xy ðtn Þ 0 0 8 9 8 0 9 " > "*ðt Þ > > > > < x n > = > < x > = "0y þ "*y ðtn Þ > > > > > > : ; > : 0 > ; 0 2"xy 9 38 x ðtr Þ > > > > = < 7 ðt Þ 0 7 y r 5> > > > nr : ðt Þ ; d66 xy r 0 Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. ðA:12Þ 542 FAZIL O. SONMEZ, H. THOMAS HAHN AND MUSTAFA AKBULUT Assuming that the Poisson’s ratio in the isotropic plane, 23, is independent of time and temperature, S23(t) is given by S23 ðtÞ ¼ 23 S22 ðtÞ ðA:13Þ Also, the following values should be used for thermal expansion coefficients: xx ¼ 11 þ 31 33 yy ¼ 22 þ 32 33 ðA:14Þ For transversely isotropic laminates 33 ¼ 22, 32 ¼ 23 and 31 ¼ 21. When the roller movement is transverse to the fiber direction for a given ply, strain–stress relations for plane strain state become 9 8 2 3 nr nr 2 nr nr nr 2 nr "x ðtn Þ > d22 ðd12 Þ =d11 23 d22 ðd12 Þ =d11 0 > > > > > n 6 = X < 7 6 7 nr nr 2 nr nr nr 2 nr "y ðtn Þ ¼ ðd12 Þ =d11 d22 ðd12 Þ =d11 0 6 23 d22 7 > > 4 5 > > r¼1 > > ; : nr 2"xy ðtn Þ 0 0 2ð1 þ 23 Þd22 9 8 9 8 0 9 8 x ðtr Þ > "*x ðtn Þ > > > "x > > > > > > > > < > > > > > > = = = < < 0 ðA:15Þ y ðtr Þ þ "*y ðtn Þ "y > > > > > > > > > > > > > > > > > > ; : ; : 0 ; : xy ðtr Þ 0 2"xy and thermal expansion coefficients are xx ¼ yy ¼ 22 þ 12 11 ðA:16Þ Equation (A.12) or (A.15) can be converted to Equation (9), where stress is a dependent variable, by following the procedure described in [15]. REFERENCES 1. Kwon, Y.W. and Berner, J.M. (1997). Matrix damage of fibrous composites: effects of thermal residual stresses and layer sequences. Computers & Structures, 64(1–4): 375–382. 2. Chapman, T.J., Gillespie, J.W., Pipes, R.B., Manson, J.-A.E. and Seferis, J.C. (1990). Prediction of process-induced stresses in thermoplastic composites. Journal of Composite Materials, 24: 616–643. 3. Jeronimidis, G. and Parkyn, A.T. (1988). Residual stresses in carbon fibre-thermoplastic matrix laminates. Journal of Composite Materials, 22: 401–415. Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. Analysis of Process-Induced Residual Stresses in Tape Placement 543 4. Lawrence, W.E., Manson, J.-A.E., Seferis, J.C., Gillespie, J.W. and Pipes, R.B. (1990). Prediction of residual stress in continuous fiber semicrystalline in thermoplastic composites: a kinetic viscoelastic approach. In: Proceedings of American Society for Composites 5th Technical Conference. Holiday Inn, 300 MAC, E. Lansing, MI, June 12–14. pp. 401–414. 5. Wang, T.M., Daniel, I.M. and Gotto, J.T. (1992). Thermoviscoelastic analysis of residual stresses and warpage in composite laminates. Journal of Composite Materials, 26(6): 883–899. 6. Unger, W.J. and Hansen, J.S. (1993). The effect of cooling rate and annealing on residual stress development in graphite fibre reinforced PEEK laminates. Journal of Composite Materials, 27(2): 108–137. 7. Lee, K. and Weistman, Y. (1994). Optimal cool down in nonlinear thermoviscoelasticity with application to graphite/PEEK (APC-2) laminates. Journal of Applied Mechanics, 61: 367–374. 8. Li, M.C., Wu, J.J., Loos, A.C. and Morton, J. (1997). A plane-strain finite element model for process-induced stresses in a graphite/PEEK composite. Journal of Composite Materials, 31(3): 212–243. 9. Wang, C. and Sun, C.T. (1997). Thermoelastic behavior of PEEK thermoplastic composite during cooling from forming temperatures. Journal of Composite Materials, 31(22): 2230–2248. 10. Domb, M.M. and Hansen, J.S. (1998). The effect of cooling rate on free-edge stress development in semi-crystalline thermoplastic laminates. Journal of Composite Materials, 32(4): 361–386. 11. Ersoy, N. and Vardar, O. (2000). Measurement of residual stresses in layered composites by compliance method. Journal of Composite Materials, 34(7): 575–598. 12. Ersoy, N.B. (1998). Prediction and measurement of residual stresses in layered composites. PhD Thesis, Bogazici University, Istanbul. 13. Nejhad, M.N.G., Gillespie, J.W. and Cope, Jr. R.D. (1992). Prediction of process-induced stresses for in-situ thermoplastic filament winding of cylinders. In: Proceedings of 3rd International Conference, Computer Aided Design in Composite Material Technology. Delaware: Newark. Vol. 3. pp. 277–295. 14. Sonmez, F.O. and Hahn, H.T. (1997). Process modeling of heat transfer and crystallization for thermoplastic composite tape placement. Journal of Thermoplastic Composite Materials, 10(3): 198–240. 15. Sonmez, F.O. and Hahn, H.T. (July 1997). Thermoviscoelastic analysis of the tape placement process. Journal of Thermoplastic Composite Materials, 10(4): 381–414. 16. Xiao, X.R. (1994). Characterization and modeling of nonlinear viscoelastic response of PEEK resin and PEEK composites. Composites Engineering, 4(7): 681–702. 17. Horoschenkoff, A. (1990). Characterization of the creep compliances J22 and J66 of orthotropic composites with PEEK and epoxy matrices using the nonlinear viscoelastic response of the neat resins. Journal of Composite Materials, 24: 879–891. 18. Morland, L.W. and Lee, E.H. (1960). Stress analysis for linear viscoelastic materials with temperature variation. Transactions of the Society of Rheology, 4: 233–263. 19. Shapery, R.A. (1967). Stress analysis of viscoelastic composite materials. Composite Materials, 1: 228–267. 20. Christensen, R.M. (1982). Theory of Viscoelasticity. New York: Academic Press. 21. Yang, T.Y. (1986). Finite Element Structural Analysis. New Jersey: Prentice-Hall. 22. Lynch, F. de S. (1969). A finite element method of viscoelastic stress analysis with application to rolling contact problems. International Journal for Numerical Methods in Engineering, 1: 379–394. 23. Batra, R.C. (1977). Cold sheet rolling, the thermoviscoelastic problem. A numerical, solution. International Journal for Numerical Methods in Engineering, 11: 671–682. Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. 544 FAZIL O. SONMEZ, H. THOMAS HAHN AND MUSTAFA AKBULUT 24. Sonmez, F.O. (1995). Modeling of the thermoplastic composite tape placement process. PhD Thesis, UCLA, Los Angeles. 25. Lee, K. and Weistman, Y. (1992). Residual thermal stresses in graphite/PEEK (APC-2) laminates. In: Fujiwara, H., Abe, T. and Tanaka, K. (eds.), Residual Stresses-III. London: Elsevier Science Publishers Ltd. pp. 31–37. 26. Barnes, J.A., Simms, I.J., Farrow, G.J., Jackson, D., Wostenholm, G. and Yates, B. (1990). Thermal expansion behavior of thermoplastic composite materials. Journal of Thermoplastic Composite Materials, 3: 66–80. 27. Barnes, J.A. (1993). Thermal expansion behavior of thermoplastic composites. Journal of Material Science, 28: 4974–4982. 28. Manson, J.-A. and Seferis, J.C. Internal stress determination by process simulated laminates. In: SPE ANTEC 1987 Conference Proceedings. Los Angeles, CA. Downloaded from http://jtc.sagepub.com at BOGAZICI UNIV LIBRARY on September 17, 2007 © 2002 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.